Exploring Gender Disparities in Automatic Speech Recognition Technology

Hend ElGhazaly; Bahman Mirheidari; Nafise Sadat Moosavi; Heidi Christensen

Exploring Gender Disparities in Automatic Speech Recognition Technology

Computation and Language 2025-02-26 v1 Sound Audio and Speech Processing

Authors: Hend ElGhazaly , Bahman Mirheidari , Nafise Sadat Moosavi , Heidi Christensen

Abstract

This study investigates factors influencing Automatic Speech Recognition (ASR) systems' fairness and performance across genders, beyond the conventional examination of demographics. Using the LibriSpeech dataset and the Whisper small model, we analyze how performance varies across different gender representations in training data. Our findings suggest a complex interplay between the gender ratio in training data and ASR performance. Optimal fairness occurs at specific gender distributions rather than a simple 50-50 split. Furthermore, our findings suggest that factors like pitch variability can significantly affect ASR accuracy. This research contributes to a deeper understanding of biases in ASR systems, highlighting the importance of carefully curated training data in mitigating gender bias.

Keywords

automatic speech recognition fairness in machine learning speaker verification

Cite

@article{arxiv.2502.18434,
  title  = {Exploring Gender Disparities in Automatic Speech Recognition Technology},
  author = {Hend ElGhazaly and Bahman Mirheidari and Nafise Sadat Moosavi and Heidi Christensen},
  journal= {arXiv preprint arXiv:2502.18434},
  year   = {2025}
}

Comments

ISCA/ITG Workshop on Diversity in Large Speech and Language Models

Exploring Gender Disparities in Automatic Speech Recognition Technology

Abstract

Keywords

Cite

Comments

Related papers